Int J Adv Manuf Technol (2010) 47:81–97 DOI 10.1007/s00170-009-2100-1
SPECIAL ISSUE - ORIGINAL ARTICLE
Study of process parameters effect on the filling phase of micro-injection moulding using weld lines as flow markers G. Tosello & A. Gava & H. N. Hansen & G. Lucchetta
Received: 3 July 2008 / Accepted: 7 May 2009 / Published online: 28 May 2009 # Springer-Verlag London Limited 2009
Abstract Micro-injection moulding (micro-moulding) is a process which enables the mass production of polymer microproducts. In order to produce high-quality injection moulded micro-parts, a crucial aspect to be fully understood and optimised is the filling of the cavity by the molten polymer. As a result, the relationships between the filling pattern and the different process parameter settings have to be established. In this paper, a novel approach based on the use of weld lines as flow markers to trace the development of the flow front during the filling is proposed. The effects on the filling stage of process parameters such as temperature of the melt, temperature of the mould, injection speed and packing pressure have been investigated. An optical coordinate measuring machine has been employed for the investigation. The micro-cavity, which presents micro-features ranging from 600 μm down to 150 μm, has been manufactured by micro-electrodischarge machining. A commercially available polystyrene grade polymer has been moulded using a high-speed injection moulding machine. The design of experiment technique was employed to determine the effect of the process parameters on the filling phase of the micro-cavity. In addition, extensive measuring uncertainty analysis was performed to validate the experimental plan. Results show that the temperature of the mould and the injection speed are the most influencing G. Tosello (*) : H. N. Hansen Department of Mechanical Engineering, Technical University of Denmark (DTU), Produktionstorvet, Building 427S, Kongens Lyngby 2800, Denmark e-mail:
[email protected] A. Gava : G. Lucchetta Department of Innovation in Mechanics and Management (DIMEG), University of Padova, Via Venezia 1, Padova 35131, Italy
process parameters during the injection moulding of a micro-component. Keywords Micro-injection moulding . Weld lines . Design of experiments . Process analysis . Uncertainty
1 Introduction In the last few years, the market of micro-products has been constantly increasing. Micro-engineering is a key sector for the development of new technologies for micro-applications. The use of micro-systems covers very large and differentiated fields. Some examples include information technology components (reading caps for hard disc, ink jet printers nozzles, etc.), medical and biomedical devices (pacemaker, sensors, micro-fluidic systems for analysis of bio-fluids etc.), high technology products (palm-sized high-definition displays, mobile phones etc.) and motion sensors for the automotive industry, micro-components for implant devices as hearing aid systems (micro-connectors, micro-switches etc.). For the mass fabrication of micro-products, microinjection moulding (μIM) represents one of the most important manufacturing processes because it matches the capabilities of a low-cost process and the requirements of micro-products, as dimensions in the sub-millimetre range and low tolerances (in the order of few micrometres down to sub-micrometre range). Micro-injection moulded components can be divided in the following three classes, depending on their characteristic features [18]: &
Micro-injection moulded parts that weigh from a few milligrams to a fraction of a gram and have dimensions on the micrometre scale (e.g. micro-gear, microoperating pins)
82
&
&
Int J Adv Manuf Technol (2010) 47:81–97
Injection moulded parts of conventional size which have micro-structured regions or micro-features (e.g. compact discs with data pits, optical lenses with microsurface features, information carriers in very small devices such as sensor discs) Micro-precision parts that can have any dimensions but have tolerances in the micro-scale (e.g. connectors for optical fibre technology)
The miniaturisation of components leads to new problems in micro-forming processes and more attention is needed in all production steps [5]. The know-how of conventional technology cannot be transferred easily to micro-technology since the material behaviour (e.g. its rheology) changes in the micro-scale [22]. The reliable manufacture of polymer-based microcomponents is directly connected to the capability of controlling the micro-injection moulding process. A crucial step during the process is the filling of the cavity. It is important to understand the influence of the process parameters on the filling of a micro-cavity for the optimisation of the process and to obtain completely filled micro-parts complying with the specifications. Therefore, further experimental activities are needed in order to fully understand the process dynamics during the micro-moulding process. Characterisation of the filling phase during micro-injection moulding is a challenging task, mainly due to the dimensions of the cavity (typically in the sub-millimetre range and even down to a few micrometres) and the filling time of the cavity (in the order of a few tens of milliseconds). In order to overcome those challenges, an alternative new method, based on the measurement of the weld lines path on the surface of a micro-part, is proposed and investigated. The position of weld lines was affected by the processing conditions and worked as flow markers of the melt-flow development during the filling of the cavity. The influence of four process parameters (melt and mould temperatures, injection speed and packing pressure) was analysed and will be presented. The paper is structured as follows: Section 2 presents the state-of-the-art and flow pattern analysis methods applied to micro-injection moulding. In Section 3, the method based on the use of weld lines as flow marker is introduced. In Sections 4 and 5, the experiments and the results from the application of the new method are presented. In Sections 6 and 7, the filling analysis results are presented and discussed, respectively. Finally, conclusions and recommendations for future research work are included in Section 8.
2 Filling pattern tests Different approaches can be employed for the analysis of the filling stage of the micro-injection moulding process. In
the following sections, a review of results obtained during recent research works is reported. In particular, three methodologies can be applied in order to characterise filling patterns: & & &
Short shots, where the filling pattern is obtain by means of partially filled mouldings of increasing volume (see Section 2.1) Flow visualisation, used to show the progress of the melt front in the cavity during the injection phase (see Section 2.2). Length flow test, used to evaluate the filling capacity of the moulding system in terms of achievable flow length and aspect ratio (see Section 2.3).
2.1 Short shots method In conventional injection moulding (i.e. in the macrodimensional range), a common approach used to study the development of the melt flow inside the cavity is the short shots analysis. It consists on the injection of fractions of the molten polymer volume necessary to completely fill the cavity. The application of the short shots method to microinjection moulded parts has been shown to be possible when using a micro-injection moulding machine provided with an injection plunger [19]. One of the main conditions for the applicability of such method is that the resolution of the metering process (i.e. the smallest shot volume that can be injected in a controlled manner) has to be smaller than a fraction of the part which is significant to give information about intermediate stages of the filling. This condition can be fulfilled by micro-injection moulding machines having an injection unit with a plunger. On the other hand, small injection moulding machines with the conventional plastication unit with reciprocating screw cannot provide controlled short shots in the order of fraction of 1 mm3 (i.e. the typical volume of polymer micro-parts and/or micro-features). Furthermore, in conventional machines, the acceleration of the screw may not be high enough to provide the required injection speed in the very short time needed to produce micro-short shots. As a result, despite the fact that it is actually possible to injection mould microparts with conventional injection moulding machines (especially if electrically driven and capable of high injection speed), with such machines, it is not possible to produce reliable micro-short shots. The repeatability of process conditions in terms of actual speed and injection pressure at the very beginning of the screw movement is lower than when it has reached a steady-state injection movement. Moreover, the produced incomplete micro-parts present free surfaces with a deformation due to stress relaxation and thermal contraction. This causes an approximation on the dimensional accuracy of the determination
Int J Adv Manuf Technol (2010) 47:81–97
of the actual flow front during the filling, especially if the target is accuracy in the micro-metre range. However, the short shots method has been proven to be a feasible technique for a qualitative representation of the evolution of the filling stage when performing micro-injection moulding [10, 19]. 2.2 Flow visualisation tests Flow visualisation can also be used to describe the advancement of the flow front into the cavity during the filling stage. It consists of the use of a high-speed camera capable of actually recording at high-frame rates (in the order of 103–104 frames/s) the flow advancement inside the micro-cavity. In order to achieve such results, the mould has to be provided with a lateral opening (camera access to the mould) and one side of the cavity made of a transparent material. By subsequent image processing of the recorded film of the cavity filling, it is possible to perform a timedependent analysis of the displacement of the melt flow front [4, 21]. The flow visualisation method offers a better resolution than the short shots method. Furthermore, it can be applied when moulding polymer micro-components with both micro-injection moulding machines and conventional reciprocating-screw machines (the high-time resolution is provided by the high-speed camera). On the other hand, the construction of the mould itself is quite complicated due to the presence of a perfectly aligned optical glass and an optical mirror conveying the image from the cavity to the external camera. As a consequence, the method appears to be of difficult implementation in an industrial environment. 2.3 Length flow tests Length flow tests are used to evaluate the filling capability of the moulding system in terms of flow length in the cavity and aspect ratio. Usually, a test cavity having constant cross section and a dominant dimension parallel to the flow front
83
advancing direction is used for the purpose; typically, aspect ratios above 10 are desired [13]. Downscaling is commonly employed on conventional injection moulding and has been proposed for the investigation of filling behaviour and processability during micro-injection moulding of semi-crystalline polymers (polypropylene and polyoxymethylene) as well as amorphous polymer (acrylonitrile butadiene styrene). Investigations are first carried out by moulding under different process parameters affecting the replication capabilities and the filling performance of the process. Then, the achieved flow lengths in miniaturised channels, defined as the actual length reached by the melt during the moulding, are determined and compared in order to establish the relation between flow lengths and process parameter settings.
3 Filling analysis in μIM using weld lines as flow markers In injection moulding when two or more flow fronts meet during the filling of cavities, an imperfection observable as a line is created. This defect of injection moulded parts is referred to as a weld line. Weld lines are influenced by material composition, mould design and process conditions [20]. Particularly related to mould and part design, the basic situations that conduce to the weld lines formation are the presence of [2]: & & &
Insert in the cavity (see Fig. 1) Two or more gates for the part filling (see Fig. 2) Features in the mould (e.g. pins or holes all through the thickness of the cavity; see Figs. 2 and 3)
Weld lines are visible on the surface of the part (their depth was measured with an atomic force microscope and found to be in the range between 500 to 1,500 nm [17]) and they are a clear trace of the development of the flow melt during the filling of the cavity (see Fig. 3, right). In the
Fig. 1 Weld lines formation due to the presence of an insert in the cavity during insert moulding (a, b) and complete insert moulded part (c) [14]
84
Int J Adv Manuf Technol (2010) 47:81–97
Fig. 2 Weld lines formation in a polymer micro-product produced by injection moulding due to multi-gating design and through feature in the cavity (holes diameter is 300 μm) [16]
following, the analysis of the filling stage in micro-injection moulding using the weld lines as flow markers is presented.
injection moulding process, respectively. Sections 4.4 and 4.5 deal with the dimensional measurement of the weld lines path and the measuring uncertainty estimation, respectively.
4 Experimental 4.1 Micro-cavity design and manufacture In this paragraph, the experimental implementation and investigation for the analysis of the filling of microinjection moulded parts are presented. In particular, Section 4.1 describes the design and the manufacturing of the micro-cavity; Sections 4.2 and 4.3 present the design of the experiments technique and its implementation on the
A micro-cavity has been manufactured by microelectrodischarge machining (μEDM) with a special design specifically developed to cause the formation of weld lines (see Fig. 4). The main objective of the micro-part design was to create an effective response variable related to the
500µm time Fig. 3 Simulation of the formation of a weld line due to the presence of a micro-feature (width=200 μm) in the cavity (left). Scanning electron microscope image (right) of the actual weld line as showed in the simulation
Int J Adv Manuf Technol (2010) 47:81–97
85
Fig. 4 Micro-part design (right) and micro-cavity machined by μEDM (left)
4.5 0.20 0.15 0.6 0.3
0.45 6.5
0.6
0.15
0.20
0.20 0.25
Injection point
weld line formation. The micro-EDM process allowed for the machining of accurate features in the micro-dimensional range: channels with widths of 150, 200, 300, 450 and 600 μm, respectively, were obtained. The thickness of the part was 250 μm. The plateau placed in the middle of the cavity had a thickness of 150 μm. Average surface roughness (Ra) was measured to be 0.15 μm. 4.2 Design of experiments The design of experiment (DOE) procedure is used to systematically investigate process or product variables that influence the quality of products. Factorial design is frequently used in injection moulding experiments involving several factors where it is necessary to study single factors effect and joint effect of the factors on a response (i.e. main effect and interactions) [12, 23]. In this research, design of experiments was employed to characterise the weld lines by the methodical variation of processing conditions. A full-factorial design of experiments was implemented in order to consider all kinds of main effects and interactions [11]. In particular, the two-level full-factorial design method was used in the DOE studies. The DOE procedure consisted of the following three steps: & & &
Planning and implementation: definition of the input/ output variables and development of the experimental plan Analysis: individuation of the parameters which mostly affect the of weld lines positions Measuring uncertainty implementation: identification and estimation of sources of uncertainty and analysis of DOE results
4.3 Experimental micro-injection moulding Four different process parameters were varied in order to determine their influence on the micro-injection moulding process: & & & &
Melt temperature (Tmelt) Mould temperature (Tmould) Injection speed (Inj.Speed) Packing pressure (Ppack)
A four-factor two-level full-factorial design has been carried out performing 24 =16 moulding experiments (each factor being varied between two levels; see Table 1 for the factors levels and Table 2 for the experimental design plan). In this way, the influence of the main effects as well as of the interactions was evaluated. Five replications (i.e. five injection moulded parts, see Fig. 5) were produced for each of the 16 moulding experiments. Parts were randomly selected from the production batch after a stable process was obtained. The employed machine for the injection moulding experiments was a conventional high-speed injection moulding machine (Ferromatik Milacron K60) provided with a reciprocating
Table 1 Process parameter settings Process parameters Tmelt (°C) Tmould (°C) Inj.Speed (mm/s) Ppack (hydraulic; bar)
Low level (−1)
High level (+1)
240 45 200 10
270 70 350 100
86
Int J Adv Manuf Technol (2010) 47:81–97
Table 2 Four-factor two-level full-factorial design of experiments employed in the investigation Tmelt
Tmould
Ppack
Inj.Speed
1 2 3 4 5 6 7 8 9 10 11 12 13
1 1 1 1 −1 1 1 −1 1 −1 −1 1 −1
1 1 1 1 −1 −1 −1 −1 −1 −1 −1 −1 1
1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1
1 1 −1 −1 1 1 −1 −1 1 1 −1 −1 1
14 15 16
−1 −1 −1
1 1 1
−1 1 −1
1 −1 −1
Run
screw with a diameter of 32 mm and a clamping force of 500 kN. The levels of the factors were established with the aim of keeping a realistic industry relevant perspective. In particular: & & &
&
Tmould was controlled using temperature sensors placed in the mould and set according to specifications given by the material supplier. Tmelt was implemented as barrel temperature. Upper and lower levels were set taking into account recommended values published by the material supplier. The minimum level of Inj.Speed was set to assure the complete filling of the cavity, whereas the maximum level was set taking into account the machine’s capability and set to 70% of the maximum injection speed limit. Ppack was set as 8.5% and 85% of the maximum hydraulic pressure (120 bar hydraulic, corresponding to 180 MPa at injection location).
Fig. 5 Sprue, runners and part (left); detailed view of the injection moulded part (right)
Polystyrene (BASF PS143E) was employed as polymer material. Polystyrene is a relevant polymer in microinjection moulding for its very high flowability, good biocompatibility and high transparency. 4.4 Weld lines path measurements The characterisation of the weld lines was carried out by means of two-dimensional measurements for the geometrical characterisation of their shape on the surface of the micro-injection moulded components. Position and orientation of the weld lines have been investigated by means of an optical coordinate measuring machine (CMM; see Fig. 6, left). To describe the shape of the weld lines, an accurate and repeatable measurement of the weld lines path on the surface sample was required. This was obtained by using both a repeatable alignment strategy of the workpiece in the measuring workspace of the machine (see Fig. 6, right) and by defining a repeatable local alignment systems integral with the considered micro-features (see Fig. 7). Optical measurements of points located on the weld line paths were executed by collecting an X coordinate measured at a pre-determined value of the Y coordinate (see Fig. 7, right). Hence, it has been possible to attribute numerical values to position (X and Y coordinates, see Fig. 8) of the weld lines and then relate the measurements to the level of the process parameters in a subsequent analysis step. 4.5 Measuring uncertainty An uncertainty assessment has been carried out in order to verify the quality of the measurements. The uncertainty of measurement (U) is a parameter, associated with the result of a measurement, which characterises the dispersion of the values that could reasonably be attributed to the measurand [6]. Depending on the considered output for the analysis, the uncertainty of the measurements can create a spread of the measurements which can partially/totally hide the effect of the experimental factor on the response. It was therefore
Int J Adv Manuf Technol (2010) 47:81–97 Fig. 6 Optical coordinate measuring machine (CMM, left) and micro-injection moulded part aligned to the coordinate reference system of the machine (right)
87
Z CMM
Z
YCMM
X Y
X CMM very important to assess the uncertainty of the measuring equipment and of the procedure. To calculate the uncertainty of a measurement, the identification of sources of uncertainty is needed. The standard uncertainty of each uncertainty contributor can be assessed by type A evaluation and type B evaluation [1, 8]. Moreover, uncertainty in a measuring process is in most cases a mix of known and unknown errors from a number of uncertainty sources contributors. Uncertainty contributors in measurement can be defined as related to [7] workpiece, equipment, measuring procedure, metrologist, characteristics definitions, environment and reference element. In the specific case presented in this paper, different measuring uncertainty sources have been taken into account following the indications given by the standardisation body and applied to the employed measuring procedure: & & & & &
Repeatability of the measuring process CMM calibration Uncertainty due to temperature variations Resolution of the CMM Repeatability of the injection moulding process
Fig. 7 Micro-features coordinate reference system (left) and measuring strategy to determine the coordinate of the points (right)
The uncertainty assessment of the measurements executed with the optical CMM was experimentally performed. The considered contributors, the related standard uncertainties in the measurements of the micro-moulded part, as well as the standard combined uncertainty and the expanded uncertainty were calculated and they are reported in Table 3. The combined standard uncertainty was calculated applying the error propagation law, as described in [7].
5 Analysis of results The experimental results in terms of weld lines position depending on the process conditions and on the cavity geometry were analysed. Different analyses were performed and various parameters investigated: one factor at-the-time process analysis (Section 5.1), the effect of micro-channel widths on the flow (Section 5.2), statistical analysis to characterise the polymer flow behaviour through microfeatures (Section 5.3.1) and to fill polymer microcomponents (Section 5.3.2).
88
Int J Adv Manuf Technol (2010) 47:81–97
0.90 0.80
50µm
0.70
line 3
Y [mm]
0.60 0.50 0.40 0.30 0.20
line 1
line 2
0.10 0.00 -0.10 -0.30 -0.20 -0.10 0.00
0.10
0.20
0.30 0.40
0.50
X [mm]
Fig. 8 Visualisation of weld lines on the surface of the micro-moulded parts (left) and measuring result (right)
5.1 One-factor-at-the-time process analysis In order to show the effect of a single factor on the response, a one-factor-at-the-time analysis was carried out. This simple and immediate analysis was performed as follows. Firstly, a set of reference process parameter setup was chosen (e.g. the process setup combination number 8 indicated in Table 2). Secondly, only one factor at a time was considered on a different level than the one considered on the reference process parameter setup. Finally, the response of the reference setup was compared with the result of the setup with the varied factor. Therefore, the resulting number of comparisons is equal to the number of factors (see Table 4). The different paths of weld lines obtained with the different settings of process parameters are chosen as response and are shown below [15]. A variation of Tmould and of Inj.Speed produces modification of the path of the weld lines. Such position variations are clearly distinguished and larger than the calculated measurement
uncertainty. On the other hand, variation of Ppack and of Tmelt does not produce a significant change (Figs. 9–12). 5.2 Effect of the micro-channel width on the flow The position of weld lines (i.e. the development of the flow front during the filling of the micro-cavity) was found to be influenced by the geometry of the features in the cavity, in particular by their width. In the final part of the cavity, in fact, micro-channels of different width were created. By measuring the weld lines path, which represents the flow front path position at the end of filling, it was possible to relate the channels width and the distance covered by the melt flow towards the end of the cavity (defined as depth of filling) for a given configuration of the process parameters. The requirement for this analysis is that the flow front after the plateau and at the beginning of the features at the end of the cavity should be as flat as possible (i.e. speed direction parallel to Y direction of the geometry). In the first
Table 3 Mean values of standard uncertainty contributors and expanded uncertainty for weld lines measurements on both X and Y directions Uncertainty sources
Repeatability of the measuring process—u(rep) Optical CMM calibration uncertainty—u(CMM) Temperature—u(temp) Repeatability of the μIM process—u(µIM) Combined standard uncertainty—u(comb) Expanded uncertainty—U (k=2, confidence level 95%)
Standard uncertainty (ucX(i)) (X direction) [μm]
Standard uncertainty (ucY(i)) (Y direction) [μm]
2.8 0.85 0.15 5.5 6.2 12
0.3 0.85 0.15 2.6 2.8 6
Int J Adv Manuf Technol (2010) 47:81–97
89 0.9
Table 4 Experimental design plan for the one-factor at-the-time analysis
0.8 0.7
DOE setup
Tmelt
Tmould
Ppack
Inj.Speed
-1 -1 -1 1 -1
-1 -1 -1 -1 1
-1 -1 1 -1 -1
-1 1 -1 -1 -1
0.6
InjSpeed=350mm/s InjSpeed=350mm/s InjSpeed=350mm/s InjSpeed=200mm/s InjSpeed=200mm/s InjSpeed=200mm/s
Y [mm]
0.5
8 (reference) 10 11 12 16
0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
X [mm]
half of the cavity (i.e. near the gate), different flow paths are generated having different filling length depending on their position (see Fig. 13, left). To straighten the flow path, a thin plateau (thickness of 150 μm) was designed and manufactured in the central part of the component. The plateau had the effect to create a uniform flow front before entering the second half of the cavity (see Fig. 13, right). The depth of filling (i.e. the ease of the melt flow to fill micro-structures) could be analysed as function of the features width. It increased with the width of the structures to be filled. The meeting points of weld lines out of channels 200 and 300 μm wide, as well as the horizontal weld line in the channel 150 μm, showed such behaviour (see Table 5). 5.3 Polymer flow analysis Main effects and two-way interactions were considered for the statistical analysis of the design of experiments. Main effect plot and Pareto chart were employed to represent which process parameters have a relevant influence on the flow behaviour for the filling of the micro-cavity. In particular, the characterisation of the flow pattern in micro-structures and for the complete filling of miniaturised
Fig. 10 Effect of injection speed variation
parts until the end of the cavity was carried out. These two crucial aspects were analysed by considering two different outputs, as described in Section 5.3.1 (filling of microstructures) and in Section 5.3.2 (filling of micro-parts). 5.3.1 Polymer flow in micro-structures For the optimisation of the filling of narrow structures, the weld line created inside the 150-μm-wide micro-channel was considered (see feature 2 in Fig. 14). The factors that lead to a maximum variation of the Y coordinate (according to the reference system indicated in Fig. 14, Y coordinate is negative inside the micro-feature 2) will be the factors that enhance the filling capability of the process. The Pareto chart highlights the most influent factors for this output: Inj.Speed and Tmould (see Fig. 15). The effects of Inj.Speed and of Tmould extend beyond the reference line (which indicates the significance of factors at the confidence level of 0.05). The effect of Tmelt and of the interaction between Inj.Speed and Tmould are on the limit of significance. All the other interactions are not significant.
0.9
0.9 0.8 0.7 0.6
Tmold=70°C Tmold=70°C Tmold=70°C Tmold=45°C Tmold=45°C Tmold=45°C
0.8 0.7 0.6 0.5
Y [mm]
0.5
Y [mm]
Ppack=100bar Ppack=100bar Ppack=100bar Ppack=10bar Ppack=10bar Ppack=10bar
0.4 0.3
0.4 0.3
0.2
0.2
0.1
0.1
0.0
0.0
-0.1
-0.1
-0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
-0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
X [mm]
Fig. 9 Effect of mould temperature variation
X [mm]
Fig. 11 Effect of packing pressure variation
90
Int J Adv Manuf Technol (2010) 47:81–97 0.9 0.8 0.7 0.6
Tmelt=270°C Tmelt=270°C Tmelt=270°C Tmelt=240°C Tmelt=240°C Tmelt=240°C
Y [mm]
0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
X [mm]
Fig. 12 Effect of Tmelt variation
Moreover, it has to be noted that the effect of Ppack is also not significant. The effect plot of the main factors shows the actual displacement in dimensional unit of the melt flow front due to the different level of the factors. It is interesting to notice that an increase of either Inj.Speed or Tmould has the positive effect to extend the filling of the micro-feature of 46 μm (see Fig. 16). In Fig. 17, the interaction plot for the position of the weld line inside the micro-feature shows that the effect of an increase of temperature of the mould at low injection speed is higher of the effect of the same increase of Tmould at high injection speed.
The Pareto chart highlights the most influent factors for this output: Inj.Speed and Tmould (see Fig. 20). The effects of Inj. Speed and Tmould extend beyond the reference line (significant at the default level of 0.05). All the other interactions are not significant. Moreover, it has to be noted that the effects of the Ppack as well as of Tmelt are also not significant. The effect plot of the main factors shows the actual displacement in dimensional unit of the melt flow front due to the different level of the factors. It is interesting to notice that an increase of the Inj.Speed pushes upwards the melt flow front at the end of filling of 202 μm (when passing from the low to the high level). Similarly, an increase of Tmould has the positive effect to displace the meeting point towards the end of the part (i.e. to extend the filling of the micro-part) of 126 μm (see Fig. 21).
6 Design of experiment and uncertainty analysis The dimensional measurements and the quality of measurements are fundamental steps during an experimental investigation in order to obtain reliable results for analysis. In this research, the following aspects had to be considered: & &
5.3.2 Polymer flow in micro-components Variations of the process parameters levels cause changes of the position of the weld lines meeting point at the end of filling. Hence, the process conditions which improve the flow front path towards the end of the filling of the microcomponent are investigated (see Fig. 18). A particular output was considered for this purpose: the Y coordinate of the meeting point (Ymp) of the weld lines represented in Fig. 19. The main factors leading to a maximum Ymp value will be the ones enhancing the filling capability of the process.
Fig. 13 Experimental short shot showing the flow front shape at the beginning of the cavity filling (left); experimental short shot showing the flow front straightened by the 150-μm-thick plateau (right)
&
Critical dimensions of the produced micro-parts were in the sub-millimetre range. Variations of the position of the weld lines (i.e. of the flow patterns) due to different process conditions were measured to be in the order of 101 up to 102 μm depending on the location and position considered. Measuring uncertainty should be reasonably lower than the effect due to a certain factor variation.
In metrology, a tool exists to help deciding what measuring uncertainty should an instrument possess in order to be suitable for a given measuring task. The so-called golden rule of metrology states that the measurement uncertainty shall be less than 10% of the tolerance to be verified. As tolerances become smaller, the “golden rule” of metrology cannot be fulfilled anymore because the lowest level of available uncertainty is reached. Such lower level is due not only to
Int J Adv Manuf Technol (2010) 47:81–97
91
Table 5 Experimental depth of filling depending on the channel width
Y coordinate [mm]
X coordinate [mm] 0.00 0.90 0.75 0.60 0.45 0.30 0.15 0.00 -0.15 -0.30 -0.45
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50 LINE 1 LINE 2A LINE 2B LINE 2C LINE 3A LINE 3B LINE 3C
1
2
3
4
5
6
7
Channel no.
Width [µm]
Depth of filling [µm]
1
400
900
2
450
900
3
200
301
4
600
900
5
150
-22
-
6
300
710
-
7
600
900
technological limitation but it also includes the cost of the measurement. The percentage ratio between the uncertainty and the tolerance can be changed to a margin of 20%, but does not solve the problem [9]. In case of high accuracy processes, as in the field of micro-manufacturing, the aforementioned uncertainty limit is reached. Especially in the case of microparts, uncertainty should be accurately estimated in order to
End of cavity
-
verify that the instrument being used is suitable for submillimetre dimensional ranges. In the case of the design of experiments, uncertainty and dimensional measurements play an important role being the link between the process effects on the part and the numerical data to be analysed. As a consequence, a procedure for uncertainty management capable of imple-
Fig. 14 Layout of weld lines formation in the micro-moulded part and local reference system (left); SEM picture of the weld lines in the 150-μmwide feature (feature no. 2; right)
92
Int J Adv Manuf Technol (2010) 47:81–97 35
Interaction Plot for Weld Line inside 150µm wide channel
Normalized effects [%]
30
-1
1
-1
1
-1
1
25
-0.04
20
Tmelt
-0.08
15
Tmelt
-1 1
-0.12 -0.04
10 Tmould
-0.08
Tmould
-1 1
5
-0.12 0
j
V
l ou
T
m
T
j
j
t el
d
in
m
V d*
in
V lt*
l
l
ou
Tm
V d*
P
ho
-0.04
j
in
t*P
e
T
l ho
el
m
T
d
d
in
P
t*T
el
m
T
d
ld ou
l ho
m
m
*P
l ho
ld
menting the measuring uncertainty in the statistical analysis and to verify its reliability is hereby proposed. The procedure can be summarised in the following steps: Selection of five parts for each of the 16 experiments to carry out all the measurements related to the fullfactorial design of experiments (see Section 4.3) Evaluation of the measuring uncertainty by applying dedicated procedures [7, 8] (see Section 4.5) Definition of a set of parameters able to describe the shapes of the weld lines and consequent statistical analysis (see Sections 5.3 and 5.3.2) Statistical analysis applying the uncertainty of the measurement by:
& & &
○ Defining a coefficient of uncertainty (COU) ○ Analysing the influence of the uncertainty on the Pareto analysis result (both for the effects and the reference line of significance) ○ Applying the uncertainty to the main effects plot In this research, different scenarios have been investigated, relating the uncertainty value to the actual response
Tmould
Mean of Y position [mm]
-0.04
-0.12
Fig. 17 Interaction plot for the position of the weld line inside the 150-μm-wide micro-channel
of each factor considered in the statistical analysis. For this purpose, a coefficient named COU has been defined in order to decide whether the level of uncertainty was adequate to consider reliable the analysis. &
Coefficient of uncertainty U ○ COU½% ¼ jeffect j 100 ○ where:
&
&
-0.08 -1
1
-1
Ppack
1
InjVel
-0.04
U is the expanded uncertainty of the analysed output. The average expanded uncertainty from all the 16 blocks of the design of experiments (see Table 2) was considered |effect| is the absolute value of the effect of the considered term from the Pareto analysis
The COU coefficient has been then implemented into a Pareto analysis and into the main effect plots. In case of a Pareto chart of the normalised effects, uncertainty could also be normalised (U(norm,%)) and applied to each of the normalised effect of each term (E(i)). Normalised uncertainty of the effect ○ U ðnorm; %Þ ¼ PUEðiÞ 100 i ○ where: &
-0.06
-1 1
InjVel
& Main Effects Plot for Weld Line inside 150µm wide channel Tmelt
-0.08
ou
Tm
Fig. 15 Pareto analysis of the standardised effects for the analysis of the filling of the micro-channel 150 μm wide
&
Ppack
Ppack
& &
U is the expanded uncertainty of the analysed output E(i) is the effect of the ith term i indicates the term and it is comprised between 1 and 10 (four main factor and six two-way interactions for a total of ten effects)
-0.06 -0.08 -1
1
-1
1
Fig. 16 Main effect plot for the position of the weld line inside the 150-μm-wide micro-channel
The spread of results, due to the measurement uncertainty, produced variations on the value of the reference line (see Pareto analyses in Section 5.3). This was also taken into account. For each of the outputs, it was calculated a number of virtual measurements having a Gaussian
Int J Adv Manuf Technol (2010) 47:81–97
93
Fig. 18 Optical image of weld lines in area 3 of the micromoulded part
distribution with average and standard deviation equal to the actual measured value and to the combined standard uncertainty, respectively. This was implemented by creating a routine (using the Matlab® software platform) which, given the average, the standard deviation and distribution type, was able to choose randomly an amount of values from the distribution. In this way, five virtual measurements were obtained for each output of every weld line of all the workpieces of the whole factorial plan. Hence, five virtual and one real output were used to carry out six different analyses. Then, the characteristic standard deviation of the design of experiments analysis was determined. In particular, an interval of variability of the reference line was observed. Finally, all the information could be summarised in an updated Pareto chart of the normalised effects including the following information (see Figs. 22 and 23):
An investigation has been carried out to estimate the optimum COU (and therefore the uncertainty) in order, on one hand, to contain the effort in terms of measuring
0.9
Meeting point
0.7 0.6
0.3 0.2
25 20 15 10 5
t
d
Tm el
ou l
el t*T m
-0.2 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
Tm
-0.1
Pp ac k
0
ou l
0.0
X
30
In jV el
0.1
35
Tm
0.4
Y Ymp
Y [mm]
0.5
Normalized effects [%]
40
0.8
d* In jV el ho ld *In jV T el m el t*P pa ck Tm el t*I nj Ve Tm l ou ld *P pa ck
& &
6.2 COU optimisation
P
&
The effects of the terms, ordered depending of their magnitude effect The range of variability of each effect (normalised uncertainty of the effect) The percentage COU for each of the terms’ effect The variability of the reference line due to the uncertainty of measurements
The two main contributors for an optimised filling of micro-structures (Inj.Speed and Tmould) had a COU much lower (12.0% and 12.2%, respectively) than the other terms (see Fig. 22). Therefore, the reliability of the result was proven by a low experimental measuring uncertainty compared to the effect. This is also shown by the fact that the effects, taking into account the uncertainty range, does not intersect the reference line (considered with their variability due to the uncertainty). The two terms with COU just above 30% (temperature of the melt and the interaction between Tmould and Inj.Speed) are now overlapping the reference line range with their uncertainty bandwidth, resulting in a not reliable evaluation of their effect. All the others term are not significant from a statistical point of view.
Tm ou ld
&
6.1 COU and factors effect
X [mm]
Fig. 19 Weld lines layout and measuring reference system for the evaluation of the melt flow behaviour at the end of filling
Fig. 20 Pareto analysis of the standardised effects for the analysis of the Y coordinate of the meeting point of weld lines of the polymer micro-component
94
Int J Adv Manuf Technol (2010) 47:81–97 Main Effects Plot for Ymp Tmelt
Tmould
&
0.85
Mean of Ymp [mm]
0.80 0.75 0.70 0.65 -1
1
-1
1
Ppack
InjVel
0.85 0.80 0.75 0.70 0.65 -1
1
-1
1
Fig. 21 Main effect plot for the position of the meeting point of weld lines at the end of filling
equipment and procedure and, on the other, that still gives reliable results in this particular DOE analysis. A series of DOE analysis has been carried out first with results from the actual measurements (as previously described) and later on by increasing the uncertainty performing a parametric study. The effect of an increase of the measuring uncertainty U (from 12 to 40 μm) on the DOE analysis is simulated and shown in the chart depicted in Fig. 23. As a consequence of the increased uncertainty, three distinct phenomena can be observed: &
The COU of a significant factor (Tmould) increases from 11.2% to 31.1% and the effect range partially overlaps the reference line of significance (solid lines represent the range of reference line for the experimental case, dashed lines are related to the simulation). The threshold of significance indicated by the reference line increases. A lower accuracy on the measurement brings, as consequence, a loss of significance of the experiment. It could happen that significant factors have
&
35 30 25
Normalized effects [%]
their effect hidden by the scatter induced by the measurements. The range of the reference line increases at the increase of the measuring uncertainty. It makes more probable the overlap between the reference line range and the uncertainty bandwidth, with consequent loss of confidence of the experiment.
20 15 10 5
12.0
12.2
31.8
68.4
76.1
>10 2
ck pa *P
m ou ld
el m
86.7
m ou ld
pa
t*T
P
m T
T
34.6
T
ck
ck pa
el
t*P
nj Ve l t*I el
Tm
ck *In P
pa
ld ou Tm
-10 COU [%]
jV el
jV el
t
*In
m el T
Tm
ou
jV el
In
-5
ld
0
>103
>104
Fig. 22 Example of up-to-date Pareto chart of the measurements of the melt flow position in the micro-channel 150 μm wide
A threshold value of the COU of approximately 30% has been found to be the limit for a significant factor (from a conventional Pareto analysis) before having its effect (enlarged by the uncertainty) overlapping the reference line range (considered with its variability). Therefore, it could be concluded that a COU lower than 30% is a pre-requisite to fulfil in order to obtain reliable and accurate results from the DOE analysis. 6.3 Uncertainty applied to main effect plot The measuring uncertainty can also be applied to the main effect plot. This analysis clearly shows that the significant factors with low COU have their uncertainty ranges not overlapping proving the reliability of the result of the design of experiments analysis (see for example Tmould and Inj.Speed in Fig. 24).
7 Discussion The result of the statistical analysis is that the final position of the flow front at the end of filling could be extended by increasing the temperature of Tmould and Inj.Speed. On the other hand, increases of temperature of Tmelt and Ppack do not influence significantly the filling behaviour. These findings are in accordance with results obtained by analysing the behaviour of melt flow in a thin-walled micro-part (width and length = 4 mm, thickness = 200 μm) using a high-speed camera applied to the cavity (i.e. flow visualisation test) [21]. In this research, it was found that Tmould and Inj.Speed were the most effective parameters for the filling of the micro-cavity and they had a greater effected if compared with Tmelt. However, in [21], a different polystyrene grade (CHI MEI PG-22) was employed for the experiments instead of BASF 143 E, which was used in the present investigation. Despite the fact that the two materials are supplied by two different manufacturers, their rheological behaviours were verified to be very similar. In fact, the rheology data of both polymers, measured with a capillary rheometer (supplied by the respective material manufacturers), were compared on a typical viscosity vs. shear rate plot, and the curves were found to be almost identical (see Fig. 25). In conclusion, the rheological similarity between the two polymers could
Int J Adv Manuf Technol (2010) 47:81–97
95
Fig. 23 Example of up-to-date Pareto chart and uncertainty effect on the measurements of the weld lines meeting point at the end of filling
Normalized effects [%]
45 40
DOE with meas. U=12µm (experimental)
35
DOE with meas. U=40µm (simulation)
30 25 20 15 10 5
m el t
ck
T
pa
Tm
T
P
d m ou l
jV el In
ou ld *In jV el T m el t*P pa ck P ho ld *In jV el T m el t*I nj Ve Tm l ou ld *P pa ck T m el t*T m ou ld
0 -5
COU
7.1
[%]
Fig. 24 Uncertainty applied to the main effect plot of the analysis melt flow position in the micro-channel 150 μm wide
20.5
11.2 31.1
26.9 76.8
28.1 81.9
28.4
53.3
85.7
2
>10
>10
2
>10
2
2
>10
2
>10
2
>10
2
>10
2
>10
3
T mould
-0.02
-0.02
-0.03
-0.03
Flow front position [mm]
Flow front position [mm]
V. Inj.
>10
-0.04 -0.05 -0.06 -0.07
-0.04 -0.05 -0.06 -0.07 -0.08
-0.08
-0.09
-0.09 -1
Level factor
+1
-1
Level factor
+1
1.0E+03 143 E_200°C
143 E_225°C 143 E_250°C
Viscosity [Pa*s]
1.0E+02
PG-22_200°C PG-22_225°C PG-22_250°C
1.0E+01
1.0E+00
1.0E-01 1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E+06
Shear rate [1/s]
Fig. 25 Comparison of the rheological behaviour of two different polystyrene grades: BASF 143 E and CHI MEI PG-22
Fig. 26 Simulated flow front pattern to be compared with weld lines depicted in Table 5 used as flow markers [16]
96
be safely assumed, confirming the validity of the conclusions obtained with the two different filling pattern tests (i.e. flow visualisation test and flow markers test). Hence, it can be concluded that complete filling of micro-injection moulded features (i.e. a good filling pattern) can be obtained by using a high Tmould (which decreases the viscosity of the melt and prevents premature solidification) and high Inj.Speed (which also decreases the viscosity of the melt due to shear thinning and viscous heating, as well as decreasing the injection time avoiding premature short shots and incomplete filling). On the other hand, it is not convenient to increase Tmelt, firstly due to the limited benefit on the filling pattern and secondly to avoid material degradation due to material over-heating. An elevated Ppack is also not advantageous because it can produce high internal tension on the polymer matrix as well as induce high stress on the mould itself. The influence of the measuring uncertainty on the reliability of the design of experiments results was also investigated. It was observed that the necessary condition to be fulfilled, in order to obtain reliable results from the analysis of the DOE, is the uncertainty of the employed measuring process being lower than the 30% of the term effect of the most effective factors indicated by the Pareto analysis. The quality of results could be proved also by applying the measuring uncertainty to the main factor plot. It was observed that the effects produced by the variation of the significant process parameters (Inj.Speed and Tmould) were larger than the measuring uncertainty, confirming the validity of the experimental implementation and of the metrological framework.
8 Conclusion The application of a viable method to understand and characterise the filling of micro-cavity during microinjection moulding is the key step for the optimisation of the process and the complete filling of the part. A new procedure based on the use of weld lines as flow markers in a polymer micro-part has been presented. The influence of four parameters of the micro-injection moulding process on weld lines position and shape has been investigated: temperature of the melt, temperature of the mould, injection speed and packing pressure. A full-factorial design of experiments (24 =16 trials) has been implemented in order to investigate the effect of the process parameters. The produced micro-parts have been investigated by using two-dimensional measurements. An optical coordinate measuring machine has been employed for the investigation. The output considered in the analysis was the position (i.e. X and Y coordinates of points) of different weld lines paths. The statistical analysis of the design of experiments indicated the injection velocity and mould
Int J Adv Manuf Technol (2010) 47:81–97
temperatures as the process parameters having the highest influence on the filling of micro-cavities. The effect of features width on the filling pattern was also analysed and evaluated. The influence of the measuring uncertainty was investigated. It was observed that the necessary condition to be fulfilled, in order to obtain reliable results from the analysis of the DOE, is the uncertainty of the employed measuring process being lower than the 30% of the term effect of the most effective factors indicated by the Pareto analysis. Finally, future works will regard the injection moulding of the presented micro-part with other polymers having different characteristics. In particular, the investigation of the filling behaviour in micro-cavities of semi-crystalline materials (such as POM and polyoxymethylene and polyamide (PA)) is of great interest. Moreover, the effect of the presence of glass fibres or nano-fillers (nano-clays or carbon nano-tubes) on the filling pattern of the polymer blend can be determined quantitatively and compared with the pattern of the unfilled polymer. In addition, the flow marker method can be effectively employed for the validation of software simulation in terms of flow front prediction during the filling of micro-cavities. Comparison between experimental and simulated weld lines can help as term of reference for a quantitative evaluation of the accuracy of the simulation results (see Fig. 26) [3]. The solid metrology framework on which the flow marker method is based on is a clear advantage when quantitative comparisons amongst different experiments and/or simulation of filling during micro-injection moulding are considered. Acknowledgements The present research was carried out within a joint research programme between the Department of Mechanical Engineering at the Technical University of Denmark (DTU) and the Department of Innovation in Mechanics and Management (DIMEG) at University of Padova (Italy) to whom we extend our thanks. Furthermore, the collaboration from Ph.D. M. Salvador (DIMEG) for his contribution in the design phase of the project is gratefully acknowledged. M.Sc. P. Motta, M.Sc. M. Dal Molin and M.Sc. M. Guarise (at DTU) are acknowledged for their contribution in connection with the experimental implementation of the research. Ph.D. E.M. Barini (Politecnico di Torino, Italy) is also acknowledged for valuable discussion regarding design of experiments and uncertainty assessment during the completion of the paper. Also, this work was carried out within the activities of the European Network of Excellence 4M “Multi Material Micro Manufacture: Technology and Applications” (European Community founding FP6-500274-1; www.4m-net.org) and in connection with the activities of the Processing of Polymer Technology Division (4M Work Package 4; www.4m-polymermicro.org).
References 1. EA (European co-operation for Accreditation) (1999) Expression of the uncertainty of measurement in calibration, EA-4/02. EA, Paris 2. Fellahi S et al (1995) Weld lines in injection molded parts: a review. Adv Polym Technol 14(3):169–195
Int J Adv Manuf Technol (2010) 47:81–97 3. Gava A, Tosello G, Lucchetta G, Hansen HN, Salvador M. (2007) A new approach for the validation of filling simulations in micro injection moulding. 9th International Conference on Numerical Methods in Industrial Forming Processes (NUMIFORM), Porto (Portugal), 17–27 June, pp 307–312 4. Han X, Yokoi H (2006) Visualization analysis of the filling behaviour of melt into microscale V-groove during the filling stage of injection molding. Polym Eng Sci 46(11):1590–1597 5. Heckele M, Schomburg WK (2004) Review on micro molding of thermoplastic polymers. J Micromechanics Microengineering 14 (3):R1–R14 6. ISO (International Organization for Standardization) (1993) International vocabulary of basic and general terms in metrology, 2nd edn. ISO, Geneva 7. ISO (International Organization for Standardization) (1995) GUM (guide to the expression of uncertainty in measurement). ISO, Geneva 8. ISO (International Organization for Standardization) (1999) Geometrical product specifications (GPS)—inspection by measurement of workpieces and measuring equipment—part 2: guide to the estimation of uncertainty in GPS measurement, in calibration of measuring equipment and in product verification, ISO/TS 14253-2. ISO, Geneva 9. Knapp, W (2001) Tolerance and uncertainty. Fifth International Conference on Laser Metrology, Machine Tool, CMM and Robot Performance, LAMDAMAP 2001, pp 357–366 10. Mehta MN, Barry MFC, Bibber TD, Tully D (2003) Validation of flow simulation for micromolded parts. Proceedings of the 62nd Annual Technical Conference (ANTEC) of the Society of Plastic Engineers (SPE), pp 3550–3554 11. Montgomery DC (2001) Design and analysis of experiments, 5th edn. Wiley, New York 12. Park K, Ahn JH (2004) Design of experiment considering twoway interactions and its application to injection molding processes with numerical analysis. J Mater Process Technol 146:221–227
97 13. Sha B, Dimov S, Griffiths C, Packianather MS (2007) Microinjection moulding: factors affecting the achievable aspect ratios. Int J Adv Manuf Technol 33:147–156 14. Tosello G, Hansen HN, Tang PT (2006) In-process assembly of micro metal inserts in a polymer matrix. 7th International Congress Molded Interconnect Devices (MID), pp 279–292 15. Tosello G, Gava A, Hansen HN, Lucchetta G (2007) Influence of process parameters on the weld lines of a micro injection molded component. Society of Plastics Engineers—65th Annual Technical Conference (ANTEC2007), Cincinnati (Ohio, US), 6–10 May 2007, pp 2014–2018 16. Tosello G (2008) Precision moulding of polymer micro components. Ph.D. thesis, Department of Mechanical Engineering, Technical University of Denmark, pp 1–242, ISBN 978-87-89502-77-9 17. Tosello G, Gava A, Hansen HN, Lucchetta G, Marinello F (2009) Characterization and analysis of weld lines on micro injection moulded parts using atomic force microscopy (AFM). Wear 266 (5–6):534–538. doi:10.1016/j.wear.2008.04.077 Metrology and Properties of Engineering Surfaces 18. Turng L-S (2001) Special and emerging injection molding processes. J Inj Molding Technol 5(3):160–179 19. Whiteside BR, Martyn MT, Coates PD (2005) In-process monitoring of micromoulding—assessment of process variation. Int J Polym Process XXI(5):162–169 20. Wu CH, Liang WJ (2005) Effect of geometry and injectionmolding parameters on weld line strength. Polym Eng Sci 45 (7):1021–1030 21. Yang SY, Nian SC, Sun IC (2002) Flow visualization of filling process during micro-injection molding. Int Polym Process 17 (4):354–360 22. Yao D, Kim B (2004) Scaling issues in miniaturization of injection molded parts. J Manuf Sci Eng 126:733–739 23. Zhao J, Mayes RH, Chen G, Xie H, Chan PS (2003) Effects of process parameters on the micro molding process. Polym Eng Sci 43(9):1542–1554